This is an implementation of Enhanced BiLSTM Inference Model for Natural Language Inference in Keras. The model is based on a paper by Chen et al. Link : Enhanced LSTM for Natural Language Inference
Dataset used is The Stanford Natural Language Inference (SNLI) Corpus. The model uses pre-trained word vectors, GloVe: Global Vectors for Word Representation.
Keras is needed to train the model, and test it. NLI.ipynb needs to be executed for training and testing the model.
Once the model is trained, the following files are generated :
1. tokenizer.pickle - tokenizes sentences
2. embeddings.npy - Word embeddings based on the GloVe model.
3. NLI.h5 - trained weights for the EBIM model
These files are used by app.py for predicting the class of a given input.